Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Chapter 4 - Modelling

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

MECH3361/9361 Semester 2, 2016

MODELLING AND SOLUTION


4.1. Challenges in mechanics modelling
To quantify the mechanics of a physical system, we must solve for 15 unknowns:

6 stresses (𝜎𝑥𝑥 , 𝜎𝑦𝑦 , 𝜎𝑧𝑧 , 𝜎𝑥𝑦 , 𝜎𝑦𝑧 , 𝜎𝑧𝑥 )


6 strains (𝜀𝑥𝑥 , 𝜀𝑦𝑦 , 𝜀𝑧𝑧 , 𝜀𝑥𝑦 , 𝜀𝑦𝑧 , 𝜀𝑧𝑥 )
3 displacements (𝑢, 𝑣, 𝑤)

We have a number of “basic equations” to enable this endeavour:

3 equilibrium equations
𝜕𝜎𝑥𝑥 𝜕𝜎𝑥𝑦 𝜕𝜎𝑥𝑧
+ + + 𝜌𝑓𝑥 = 𝜌𝑎𝑥
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜎𝑦𝑥 𝜕𝜎𝑦𝑦 𝜕𝜎𝑦𝑧
+ + + 𝜌𝑓𝑦 = 𝜌𝑎𝑦
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜎𝑧𝑥 𝜕𝜎𝑧𝑦 𝜕𝜎𝑧𝑧
+ + + 𝜌𝑓𝑧 = 𝜌𝑎𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
6 strain-displacement equations
𝜕𝑢 𝜕𝑣 𝜕𝑤
𝜀𝑥𝑥 = , 𝜀𝑦𝑦 = , 𝜀𝑧𝑧 =
𝜕𝑥 𝜕𝑦 𝜕𝑧
1 𝜕𝑣 𝜕𝑢 1 𝜕𝑤 𝜕𝑣 1 𝜕𝑢 𝜕𝑤
𝜀𝑥𝑦 = ( + ) , 𝜀𝑦𝑧 = ( + ) , 𝜀𝑧𝑥 = ( + )
2 𝜕𝑥 𝜕𝑦 2 𝜕𝑦 𝜕𝑧 2 𝜕𝑧 𝜕𝑥
6 stress-strain equations
𝐸 𝐸
𝜎𝑥𝑥 = [(1 − 𝜈)𝜀𝑥𝑥 + 𝜈(𝜀𝑦𝑦 + 𝜀𝑧𝑧 )] 𝜎𝑥𝑦 = 𝜀
(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈) 𝑥𝑦

𝐸 𝐸
𝜎𝑦𝑦 = [(1 − 𝜈)𝜀𝑦𝑦 + 𝜈(𝜀𝑧𝑧 + 𝜀𝑥𝑥 )] 𝜎𝑦𝑧 = 𝜀
(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈) 𝑦𝑧

𝐸 𝐸
𝜎𝑧𝑧 = [(1 − 𝜈)𝜀𝑧𝑧 + 𝜈(𝜀𝑥𝑥 + 𝜀𝑦𝑦 )] 𝜎𝑧𝑥 = 𝜀
(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈) 𝑧𝑥

Additional compatibility equations


𝜕 2 𝜀𝑥𝑥 𝜕 2 𝜀𝑦𝑦 𝜕 2 𝜀𝑥𝑦
+ = 2
𝜕𝑦 2 𝜕𝑥 2 𝜕𝑥𝜕𝑦
𝜕 2 𝜀𝑦𝑦 𝜕 2 𝜀𝑧𝑧 𝜕 2 𝜀𝑦𝑧
+ = 2
𝜕𝑧 2 𝜕𝑦 2 𝜕𝑦𝜕𝑧
𝜕 2 𝜀𝑧𝑧 𝜕 2 𝜀𝑥𝑥 𝜕 2 𝜀𝑧𝑥
+ =2
𝜕𝑥 2 𝜕𝑧 2 𝜕𝑧𝜕𝑥

1
MECH3361/9361 Semester 2, 2016

There are two main challenges:

1. How do we translate the engineering problem into a corresponding mechanics model?


2. We need to solve for 15 equations simultaneously. (This is difficult.)

4.2. Boundary conditions


Let’s look at an example from manufacturing. The cutting tool in a lathe must be sufficiently
stiff to avoid vibration and tool-tip deflection. When we want to know whether the tool is
good enough for a turning operation, we need to analyse its deformation under some turning
force.

Top view
z
y Cutting
zone A

x
Cutting
tool

Fig. 4.1 Lathe and cutting tool modelling

Here’s what we know about the cutting tool. Firstly, since the cutting tool is clamped onto the
tool post or the square turret of the lathe, the clamped end is not able to move. Secondly,
some cutting force is applied in the cutting zone. Thus, the problem can be translated into the
mechanics model shown to the bottom right of Fig. 4.1.

4.2.1. Displacement and stress boundary conditions


We can define two classes of “boundary conditions” (BCs):

Displacement BCs
(e.g. at 𝑥 = 0, 𝑢 = 𝑣 = 𝑤 = 0)
Stress BCs
(e.g. at the tool-workpiece interaction zone, 𝜎𝑛𝑥 = 𝐹𝑥 , 𝜎𝑛𝑦 = 𝐹𝑦 , 𝜎𝑛𝑧 = 𝐹𝑧 )

Specifically, with regards to the stress BC, we can say that:

2
MECH3361/9361 Semester 2, 2016

𝐹𝑥 = (𝑙𝜎𝑥𝑥 + 𝑚𝜎𝑦𝑥 + 𝑛𝜎𝑧𝑥 )𝐴 0 = 𝑙𝜎𝑥𝑥 + 𝑚𝜎𝑦𝑥 + 𝑛𝜎𝑧𝑥


Inside A: {𝐹𝑦 = (𝑙𝜎𝑥𝑦 + 𝑚𝜎𝑦𝑦 + 𝑛𝜎𝑧𝑦 )𝐴 Outside A: {0 = 𝑙𝜎𝑥𝑦 + 𝑚𝜎𝑦𝑦 + 𝑛𝜎𝑧𝑦
0 = 𝑙𝜎𝑥𝑧 + 𝑚𝜎𝑦𝑧 + 𝑛𝜎𝑧𝑧
𝐹𝑧 = (𝑙𝜎𝑥𝑧 + 𝑚𝜎𝑦𝑧 + 𝑛𝜎𝑧𝑧 )𝐴

Consider Fig. 4.2 below:

Stress B.C.

Displacement
B.C.

Mixed
B.C.
Fig. 4.2 Boundary conditions

If the displacement at (𝑥 ′ , 𝑦 ′ , 𝑧 ′ ) is 𝑢 ′ , 𝑣 ′ , 𝑤′, a correct solution to 𝑢, 𝑣, 𝑤 must satisfy:

𝑢 = 𝑢 ′ , 𝑣 = 𝑣 ′ , 𝑤 = 𝑤′
′′ ′′ ′′
Similarly, if the stresses at a surface point (𝑥”, 𝑦”, 𝑧”) are known as 𝜎𝑛𝑥 , 𝜎𝑛𝑦 , 𝜎𝑛𝑧 , the correct
solution to stress functions 𝜎𝑥𝑥 , 𝜎𝑦𝑦 , 𝜎𝑧𝑧 , 𝜎𝑥𝑦 , 𝜎𝑦𝑧 , 𝜎𝑧𝑥 must satisfy the stress B.C. as:

′′
𝜎𝑛𝑥 = 𝑙𝜎𝑥𝑥 + 𝑚𝜎𝑦𝑥 + 𝑛𝜎𝑧𝑥
′′
{𝜎𝑛𝑦 = 𝑙𝜎𝑥𝑦 + 𝑚𝜎𝑦𝑦 + 𝑛𝜎𝑧𝑦
′′
𝜎𝑛𝑧 = 𝑙𝜎𝑥𝑧 + 𝑚𝜎𝑦𝑧 + 𝑛𝜎𝑧𝑧

BCs are an important mathematical requirement in solving field problems. For mechanics
modelling, we use a number of differential equations to model the underlying physics. Since
differential equations only provide relationships from one point to the next, absolute values
(for displacement, etc.) cannot be calculated unless appropriate BCs are applied.

Example 4.1
A prismatic squared beam (𝑎×𝑎×𝐿) is clamped on its sides. A heavy cubic object (𝑎×𝑎×𝑎) is
placed on top of the beam. The density of this object is 𝜌. Describe the BCs of the beam.

3
MECH3361/9361 Semester 2, 2016

Solution

For a complete answer, we must define both the surfaces and the associated BC
mathematically.

At the surfaces defined by ABCD:

−𝑎 ≤ 𝑥 ≤ 0, 𝑦 = 0, −𝑎 ≤ 𝑧 ≤ 0

and EFGH:

−𝑎 ≤ 𝑥 ≤ 0, 𝑦 = 𝐿, −𝑎 ≤ 𝑧 ≤ 0

the beam is clamped, so the displacement BC at these surfaces is:

𝑢=𝑣=𝑤=0

At surfaces defined by ABFE:

𝑥 = 0, 0 ≤ 𝑦 ≤ 𝐿, −𝑎 ≤ 𝑧 ≤ 0

and DCGH:

𝑥 = −𝑎, 0 ≤ 𝑦 ≤ 𝐿, −𝑎 ≤ 𝑧 ≤ 0

there are no stresses (free x-surfaces), so the stress BC at these surfaces is:

𝜎𝑥𝑥 = 𝜎𝑥𝑦 = 𝜎𝑥𝑧 = 0

At surfaces defined by AIJD:

−𝑎 ≤ 𝑥 ≤ 0, 0 ≤ 𝑦 ≤ 𝑏, 𝑧 = 0

KEHL:

−𝑎 ≤ 𝑥 ≤ 0, (𝑎 + 𝑏) ≤ 𝑦 ≤ 𝐿, 𝑧 = 0

and BFGC:

−𝑎 ≤ 𝑥 ≤ 0, 0 ≤ 𝑦 ≤ 𝐿, 𝑧 = −𝑎

there are no stresses (free z-surfaces), so the stress BC at these surfaces is:

𝜎𝑧𝑥 = 𝜎𝑧𝑦 = 𝜎𝑧𝑧 = 0

At the surface defined by ILKJ:

−𝑎 ≤ 𝑥 ≤ 0, 𝑏 ≤ 𝑦 ≤ (𝑎 + 𝑏), 𝑧 = 0

4
MECH3361/9361 Semester 2, 2016

there is a compressive load only:

𝜎𝑧𝑧 = −𝑔𝜌𝑎; 𝜎𝑧𝑥 = 𝜎𝑧𝑦 = 0

4.2.2. Mixed boundary conditions


At some boundaries, there can be both displacement and stress conditions. Consider, for
instance, a constrained slider as follows:

Sliding block 1
P
b
x
Slide way
b a a
2

Fig. 4.3 A mixed boundary condition

On the top and bottom surfaces, there is a displacement BC in the normal (y) direction, but
there is also a friction stress in the tangential (x) direction. Therefore, we can define the BCs
as follows:

At the top surface – 𝑎 ≤ 𝑥 ≤ 𝑎, 𝑦 = 𝑏: 𝑣 = 0, 𝜎𝑦𝑥 = 𝜏1

At the bottom surface −𝑎 ≤ 𝑥 ≤ 𝑎, 𝑦 = −𝑏: 𝑣 = 0, 𝜎𝑦𝑥 = −𝜏2

4.2.3. Special case: Modelling symmetry


We can use displacement boundary conditions to simplify symmetrical problems. For a
mirror axis to exist, both geometry and loading must be the same for any two points on
opposite sides of the axis. Such planar symmetry problems can be modelled using only half
of the original geometry.

To denote symmetry, we use the fact that no displacement can occur across the axis, since
this would violate the compatibility of strains. For the through-hole plate below, we can
reduce the problem domain because of the symmetry about both the x and y axes. Left-right
symmetry about the (vertical) y-axis implies that at 𝑥 = 0, 𝑢 = 0. Conversely, there is top-
bottom symmetry about the (horizontal) x-axis, so we know that at 𝑦 = 0, 𝑣 = 0. These
displacement restrictions are illustrated using the fixed links as shown for the quarter-plate.

5
MECH3361/9361 Semester 2, 2016

y
𝜎 = 1 MPa
C
I H
ux=0
A a
2m
B
x
uy=0

Mild Steel
F G
2m

4.3. Plane stress


4.3.1. Rationale for simplifying analyses
Since 3D problems require many more equations to be solved, it would be useful to use 2D
approximations where reasonable to increase efficiency. Two particular configurations are
commonly used in mechanics modelling: plane stress and plane strain.

4.3.2. Definition
Many engineering structures are made of thin plate with the loads applied in the plane, e.g. a
movable white board in the classroom.

F F
t h,b

h
y
mg
x
z b
Fig. 4.4 Example of plane stress

In such a problem, 𝑡 ≤ ℎ, 𝑏 and all the loads are applied in-plane as shown in Fig. 4.4. On the
two stress-free surfaces of the plate (front and back):

𝜎𝑧𝑥 = 𝜎𝑧𝑦 = 𝜎𝑧𝑧 = 0

Thus, the non-zero stress components are 𝜎𝑥𝑥 , 𝜎𝑦𝑦 , and 𝜎𝑥𝑦 . These stress components are
independent of coordinate 𝑧, i.e.

6
MECH3361/9361 Semester 2, 2016

𝜎𝑥𝑥 = 𝜎𝑥𝑥 (𝑥, 𝑦), 𝜎𝑦𝑦 = 𝜎𝑦𝑦 (𝑥, 𝑦), 𝜎𝑥𝑦 = 𝜎𝑥𝑦 (𝑥, 𝑦)

or in other words, all of the non-vanishing stresses are in the xy-plane only. We therefore call
these “plane stress” problems.

From generalised Hooke’s law, we can see that all strains components are also independent
of 𝑧:

𝜀𝑥𝑥 = 𝜀𝑥𝑥 (𝑥, 𝑦), 𝜀𝑦𝑦 = 𝜀𝑦𝑦 (𝑥, 𝑦), 𝜀𝑥𝑦 = 𝜀𝑥𝑦 (𝑥, 𝑦)

𝜀𝑧𝑥 = 𝜀𝑧𝑦 = 0

1 𝜈
𝜀𝑧𝑧 = [𝜎𝑧𝑧 − 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 )] = − (𝜎𝑥𝑥 + 𝜎𝑦𝑦 )
𝐸 𝐸

Note that 𝜀𝑧𝑧 ≠ 0, but it is not an independent variable (it can be derived from 𝜎𝑥𝑥 and 𝜎𝑦𝑦 ).

Recalling the strain-displacement equation, we can define the displacements as:

𝑢 = 𝑢(𝑥, 𝑦) and 𝑣 = 𝑣(𝑥, 𝑦)

Again however, note that 𝑤 is not an independent variable since

𝜕𝑤 𝜈
= 𝜀𝑧𝑧 = − (𝜎𝑥𝑥 + 𝜎𝑦𝑦 )
𝜕𝑧 𝐸

4.3.3. Independent variables


All the non-vanishing and independent variables are:

Stress: 𝜎𝑥𝑥 = 𝜎𝑥𝑥 (𝑥, 𝑦), 𝜎𝑦𝑦 = 𝜎𝑦𝑦 (𝑥, 𝑦), 𝜎𝑥𝑦 = 𝜎𝑥𝑦 (𝑥, 𝑦)
Strain: 𝜀𝑥𝑥 = 𝜀𝑥𝑥 (𝑥, 𝑦), 𝜀𝑦𝑦 = 𝜀𝑦𝑦 (𝑥, 𝑦), 𝜀𝑥𝑦 = 𝜀𝑥𝑦 (𝑥, 𝑦)
Displacements: 𝑢, 𝑣

Therefore, we can simplify our mechanics equations as follows:

𝜕𝜎𝑥𝑥 𝜕𝜎𝑥𝑦
+ + 𝑏𝑥 = 𝜌𝑎𝑥
𝜕𝑥 𝜕𝑦
Equilibrium: {𝜕𝜎 𝜕𝜎𝑦𝑦
𝑦𝑥
+ + 𝑏𝑦 = 𝜌𝑎𝑦
𝜕𝑥 𝜕𝑦
𝜕𝑢
𝜀𝑥𝑥 = 𝜕𝑥
𝜕𝑣
Strain-displacement: 𝜀𝑦𝑦 = 𝜕𝑦
1 𝜕𝑢 𝜕𝑣
{𝜀𝑥𝑦 = 2 (𝜕𝑦 + 𝜕𝑥)
𝐸𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜈𝜎𝑦𝑦
Stress-strain relationship: {𝐸𝜀𝑦𝑦 = 𝜎𝑦𝑦 − 𝜈𝜎𝑥𝑥
𝐸𝜀𝑥𝑦 = (1 + 𝜈)𝜎𝑥𝑦

7
MECH3361/9361 Semester 2, 2016

𝜕2 𝜀𝑥𝑥 𝜕2 𝜀𝑦𝑦 𝜕2𝜀𝑥𝑦


Compatibility: + = 2 𝜕𝑥𝜕𝑦
𝜕𝑦 2 𝜕𝑥 2

4.4. Plane strain


4.4.1. Definition
In contrast to the thin plate under plane stress, we often need to analyse a prismatic structure
that is very long, say in the z-direction, where both the cross-section in this longitudinal
direction and the stress loading are uniform (unchanging). Examples include dams and rolling
processes, as shown in Fig. 4.5.

In this case, the strains in the z-direction (longitudinal) are effectively zero (since that
dimension is much longer than the other two):

𝜀𝑧𝑥 = 𝜀𝑧𝑦 = 𝜀𝑧𝑧 = 0

which implies that the displacement in the z-direction (𝑤) vanishes. Also, since the loading is
uniform, the stresses, strains, and displacements in any two cross-sections parallel to the
xy-plane must be identical. This means that once again, all the non-vanishing stresses,
strains, and displacements are functions of x and y, and all lie within the xy-plane, i.e.

𝜀𝑥𝑥 = 𝜀𝑥𝑥 (𝑥, 𝑦), 𝜀𝑦𝑦 = 𝜀𝑦𝑦 (𝑥, 𝑦), 𝜀𝑥𝑦 = 𝜀𝑥𝑦 (𝑥, 𝑦)

This situation is named “plane strain”.

4.4.2. Simplifying equations


From generalised Hooke’s law, we have:

1 1+𝜈
𝜀𝑥𝑥 = [𝜎 − 𝜈(𝜎𝑦𝑦 + 𝜎𝑧𝑧 )] 𝜀𝑥𝑦 = 𝜎𝑥𝑦
𝐸 𝑥𝑥 𝐸

1 1+𝜈
𝜀𝑦𝑦 = [𝜎 − 𝜈(𝜎𝑧𝑧 + 𝜎𝑥𝑥 )] 𝜀𝑦𝑧 = 𝜎𝑦𝑧
𝐸 𝑦𝑦 𝐸

1 1+𝜈
𝜀𝑧𝑧 = [𝜎 − 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 )] 𝜀𝑧𝑥 = 𝜎𝑧𝑥
𝐸 𝑧𝑧 𝐸

Since 𝜀𝑧𝑧 = 0, we can show that 𝜎𝑧𝑧 = 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 ). Therefore:

8
MECH3361/9361 Semester 2, 2016

1
𝜀𝑥𝑥 = [𝜎 − 𝜈(𝜎𝑦𝑦 + 𝜎𝑧𝑧 )]
𝐸 𝑥𝑥
1
= [𝜎𝑥𝑥 − 𝜈 (𝜎𝑦𝑦 + 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 ))]
𝐸
1
= [(1 − 𝜈 2 )𝜎𝑥𝑥 − 𝜈(1 + 𝜈)𝜎𝑦𝑦 ]
𝐸
𝐸
Multiplying by 1−𝜈 2, we find that:

𝐸 𝜈(1 + 𝜈) 𝜈(1 + 𝜈) 𝜈
𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜎𝑦𝑦 = 𝜎𝑥𝑥 − 𝜎𝑦𝑦 = 𝜎𝑥𝑥 − ( )𝜎
1−𝜈 2 1−𝜈 2 (1 + 𝜈)(1 − 𝜈) 1 − 𝜈 𝑦𝑦

Similarly,

𝐸 𝜈
𝜀𝑦𝑦 = 𝜎𝑦𝑦 − ( )𝜎
1 − 𝜈2 1 − 𝜈 𝑥𝑥

In shear:

𝐸 1 − 𝜈2
𝜎𝑥𝑦 = 𝜀𝑥𝑦 = 𝜎
(1 + 𝜈) 𝐸(1 − 𝜈) 𝑥𝑦

𝐸 1 1+𝜈−𝜈 𝜈
∴ 𝜀𝑥𝑦 = 𝜎𝑥𝑦 = 𝜎𝑥𝑦 = [1 + ]𝜎
1−𝜈 2 1−𝜈 1−𝜈 1 − 𝜈 𝑥𝑦

4.5. Comparing plane stress and plane strain


4.5.1. Key similarities
In both plane stress and plane strain problems:

deformation occurs in a single plane (xy, using the prior conventions)


all independent stresses, strains, and displacements are functions of coordinates x and
y only.

4.5.2. Key differences


Plane stress concerns a thin plate with stress-free surfaces normal to the z-axis. In this case:

𝜎𝑧𝑧 = 0
{ 𝑤≠0
𝜈
𝜀𝑧𝑧 = − (𝜎𝑥𝑥 + 𝜎𝑦𝑦 ) ≠ 0
𝐸

Plane strain concerns a very long prismatic component. In this case:

𝜀𝑧𝑧 = 0
{ 𝑤=0
𝜎𝑧𝑧 = 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 ) ≠ 0

9
MECH3361/9361 Semester 2, 2016

4.5.3. Hooke’s law


Observe the differences between plane stress and plane strain:

Component Plane stress Plane strain

𝝈𝒛𝒛 0 𝜎𝑧𝑧 = 𝜈(𝜎𝑥𝑥 + 𝜎𝑦𝑦 ) ≠ 0

𝜈
𝜺𝒛𝒛 𝜀𝑧𝑧 = − (𝜎 + 𝜎𝑦𝑦 ) ≠ 0 0
𝐸 𝑥𝑥

𝐸 𝜈
xx 𝐸𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜈𝜎𝑦𝑦 𝜀𝑥𝑥 = 𝜎𝑥𝑥 − ( )𝜎
1−𝜈 2 1 − 𝜈 𝑦𝑦

𝐸 𝜈
yy 𝐸𝜀𝑦𝑦 = 𝜎𝑦𝑦 − 𝜈𝜎𝑥𝑥 𝜀𝑦𝑦 = 𝜎𝑦𝑦 − ( )𝜎
1 − 𝜈2 1 − 𝜈 𝑥𝑥

𝐸 𝜈
xy 𝐸𝜀𝑥𝑦 = (1 + 𝜈)𝜎𝑥𝑦 𝜀𝑥𝑦 = [1 + ]𝜎
1−𝜈 2 1 − 𝜈 𝑥𝑦
𝐸
E* 𝐸∗ = 𝐸 𝐸∗ =
1 − 𝜈2
𝜈
𝝂∗ 𝜈∗ = 𝜈 𝜈∗ =
1−𝜈

𝐸 ∗ 𝜀𝑥𝑥 = 𝜎𝑥𝑥 − 𝜈 ∗ 𝜎𝑦𝑦


Unified
𝐸 ∗ 𝜀𝑦𝑦 = 𝜎𝑦𝑦 − 𝜈 ∗ 𝜎𝑥𝑥
Hooke’s law
𝐸 ∗ 𝜀𝑥𝑦 = (1 + 𝜈 ∗ )𝜎𝑥𝑦

4.5.4. Replacing solution methods


Unified Hooke’s law above indicates that we can easily obtain the solution to one class of
problem when given the solution for the other type.

From plane stress to plane strain solution

If we have obtained the solution to a plane stress problem, the corresponding plane strain
𝐸 𝜈
solution can be found by simply replacing the elastic constants E and with 1−𝜈2 and 1−𝜈,
respectively.

From plane strain to plane stress solution

Under plane strain conditions, we know that for Poisson’s ratio:


𝜈
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 =
1−𝜈

𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 (1 − 𝜈) = 𝜈

10
MECH3361/9361 Semester 2, 2016

∗ ∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 − 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 𝜈=𝜈

∗ ∗
𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 = (1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 )𝜈


𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
∴𝜈= ∗
1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛

For Young’s modulus:


𝐸
𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 =
1 − 𝜈2

𝐸 = 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 (1 − 𝜈 2 )
∗ 2

𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
= 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 [1 − ( ∗ ) ]
1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
∗ 2 ∗ 2

(1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 ) 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
= 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 [ 2 − 2]
∗ ∗
(1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 ) (1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 )


1 + 2𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛
= 𝐸𝑝𝑠𝑡𝑟𝑎𝑖𝑛 [ 2]

(1 + 𝜈𝑝𝑠𝑡𝑟𝑎𝑖𝑛 )

𝐸(1+2𝜈)
Therefore, by replacing the E and in the solution of a plane-strain problem by (1+𝜈)2
and
𝜈
, we can obtain the solution to the plane stress problem.
1+𝜈

Example 4.2
A “thin circular plate” of radius R is subjected to a uniform pressure 𝑝1 on its edge. The two
planar plate surfaces are free, as illustrated in the figure below.

(1−𝜈)
The in-plane (xy-plane) displacements in the plate are found to be 𝑢 = −𝑝1 𝑥 and
𝐸
(1−𝜈)
𝑣 = −𝑝1 𝑦. Find the corresponding displacements in a “long circular bar” under a
𝐸
uniform pressure 𝑝2 . Assume that the length of the bar is much longer than its radius R.

11
MECH3361/9361 Semester 2, 2016

Solution
𝐸
From the plane stress solution to a plane strain solution: Using 𝐸 ∗ = 1−𝜈2 to replace 𝐸, 𝜈 ∗ =
𝜈
1−𝜈
to replace 𝜈, and pressure 𝑝2 to replace 𝑝1 :

𝜈
(1 − 𝜈) (1 − 1 − 𝜈) (1 − 𝜈 2 ) − 𝜈(1 + 𝜈)
𝑢 = −𝑝2 𝑥 = −𝑝2 𝐸
𝑥 = −𝑝2 𝑥
𝐸 𝐸
1 − 𝜈2
(1 + 𝜈)(1 − 2𝜈)
= −𝑝2 𝑥
𝐸

Similarly,

(1 + 𝜈)(1 − 2𝜈)
𝑣 = −𝑝2 𝑦
𝐸

4.6. The principle of superposition


Complex problems can often be broken down into a number of simpler problems. We can do
this if materials and deformation are linear and the problem is independent of loading
history. This idea was used in Mechanics of Solids I—for example, we can resolve a
combined loading problem with axial tension, torsion, and bending by considering each load
separately.

y
y

x x

Fig. 4.6 Superposition principle

Integration

Solution under Solution under


concentrated force distributed force
Fig. 4.7 A foundation under arbitrarily distributed normal pressure.

12
MECH3361/9361 Semester 2, 2016

Example 4.3
For a thin elastic plate under a plane stress scenario, the stress functions for in-plane bending
are known to be 𝜎𝑦𝑦 = 𝜎𝑥𝑦 = 0 and 𝜎𝑥𝑥 = 6𝑎𝑦. For the same plate under in-plane shearing,
the stress functions are 𝜎𝑥𝑥 = 𝜎𝑦𝑦 = 0 and 𝜎𝑥𝑦 = −𝑎. Find the stress in the plate under the
combined loading scenario.

Solution

The stress functions for the two separate loading cases are known and can be illustrated as
below:

-a -a
6ay
x
6ay
y

To get the combined loading, the stress functions can be added directly as per the principle of
superposition. As such:

𝜎𝑥𝑥 = 𝜎𝑥𝑥,1 + 𝜎𝑥𝑥,2 = 6𝑎𝑦 + 0 = 6𝑎𝑦

𝜎𝑦𝑦 = 𝜎𝑦𝑦,1 + 𝜎𝑦𝑦,2 = 0 + 0 = 0

𝜎𝑥𝑦 = 𝜎𝑥𝑦,1 + 𝜎𝑥𝑦,2 = 0 − 𝑎 = −𝑎

4.7. Solution approaches


4.7.1. Overview
After defining the boundary conditions, one must solve for three sets of unknowns. Recall
from the start of the chapter that these are:

6 stresses (𝜎𝑥𝑥 , 𝜎𝑦𝑦 , 𝜎𝑧𝑧 , 𝜎𝑥𝑦 , 𝜎𝑦𝑧 , 𝜎𝑧𝑥 )


6 strains (𝜀𝑥𝑥 , 𝜀𝑦𝑦 , 𝜀𝑧𝑧 , 𝜀𝑥𝑦 , 𝜀𝑦𝑧 , 𝜀𝑧𝑥 )
3 displacements (𝑢, 𝑣, 𝑤)

It is impossible to solve for all of these together. We therefore have to solve one or two
groups first, based on what is known about the problem. Generally speaking, there are four
different methods: the displacement method, stress method, strain method, and mixed
method.

4.7.2. The displacement method


Unknowns: 𝑢, 𝑣, 𝑤

Procedure: The other two sets of unknown variables must first be eliminated from the
equations. This is done by replacing stresses and strains with displacements as follows.

13
MECH3361/9361 Semester 2, 2016

We can derive the following (see tutorial week 5, question 3):

𝜕𝐼1𝜀
(𝜆 + 𝜇) + 𝜇∇2 𝑢 + 𝜌𝑓𝑥 = 𝜌𝑎𝑥
𝜕𝑥
𝜕𝐼1𝜀
(𝜆 + 𝜇) + 𝜇∇2 𝑣 + 𝜌𝑓𝑦 = 𝜌𝑎𝑦
𝜕𝑦
𝜕𝐼 𝜀
(𝜆 + 𝜇) 1 + 𝜇∇2 𝑤 + 𝜌𝑓𝑧 = 𝜌𝑎𝑧
{ 𝜕𝑥
𝜕2 𝜕2 𝜕2 𝜕𝑢 𝜕𝑣 𝜕𝑤
where the Laplace operator ∇2 = 𝜕𝑥 2 + 𝜕𝑦 2 + 𝜕𝑧 2 and 𝐼1𝜀 = 𝜕𝑥 + 𝜕𝑦 + .
𝜕𝑧

After obtaining 𝑢, 𝑣, 𝑤, one can calculate strain by using the strain-displacement equations,
and stress by using Hooke’s law. Note that the solution must satisfy the boundary conditions.

4.7.3. The stress method


Unknowns: 𝜎𝑥𝑥 , 𝜎𝑦𝑦 , 𝜎𝑧𝑧 , 𝜎𝑥𝑦 , 𝜎𝑦𝑧 , 𝜎𝑧𝑥

Procedure: Solve for the stress component first, then strains and displacements.

4.7.4. The strain method


Unknowns: 𝜀𝑥𝑥 , 𝜀𝑦𝑦 , 𝜀𝑧𝑧 , 𝜀𝑥𝑦 , 𝜀𝑦𝑧 , 𝜀𝑧𝑥

Procedure: Solve for the strain component first, then stresses and displacements.

4.8. Solution to cylinders under internal and external pressure


4.8.1. Introduction
Cylindrical pressure vessels are a common occurrence in engineering. They are typically
illustrated with section diagrams as below:

po r

pi Ri
z
Ro

Fig. 4.8 Pressurised cylinder

Given the cylindrical geometry of these vessels, it is convenient to use the cylindrical
coordinate system introduced earlier for analysis. Like the Cartesian system, the cylindrical
system consists of 3 independent coordinates—in this case, (𝑟, 𝜃, 𝑧). The stress tensor for this
system has the following layout.

14
MECH3361/9361 Semester 2, 2016

dr
𝜎𝑟𝑟 𝜎𝑟𝜃 𝜎𝑟𝑧 dz

𝜎(𝑟, 𝜃, 𝑧) = [𝜎𝜃𝑟 𝜎𝜃𝜃 𝜎𝜃𝑧 ]


d
𝜎𝑧𝑟 𝜎𝑧𝜃 𝜎𝑧𝑧 dV=rd drdz
From cylindrical to Cartesian
dx=dr
dy=rd
dz=dz

Fig. 4.9 Cylindrical coordinate system and stress tensor

Our mechanics equations will take on slightly different forms.

Equilibrium equations:

In 3D, we can derive the following by replacing the coordinate system:

𝜕𝜎𝑟𝑟 1 𝜕𝜎𝑟𝜃 ∂𝜎rz 𝜎𝑟𝑟 − 𝜎𝜃𝜃


+ + + + 𝜌𝑓𝑟 = 𝜌𝑎𝑟
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

𝜕𝜎𝜃𝑟 1 𝜕𝜎𝜃𝜃 ∂𝜎𝜃z 2𝜎𝑟𝜃


+ + + + 𝜌𝑓𝜃 = 𝜌𝑎𝜃
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

𝜕𝜎𝑧𝑟 1 𝜕𝜎𝑧𝜃 ∂𝜎zz 𝜎𝑟𝑧


+ + + + 𝜌𝑓𝑧 = 𝜌𝑎𝑧
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

In 2D, 𝜎𝑟𝑧 = 𝜎𝜃𝑧 = 𝜎𝑧𝑧 = 0, therefore the equations simplify to

𝜕𝜎𝑟𝑟 1 𝜕𝜎𝑟𝜃 𝜎𝑟𝑟 − 𝜎𝜃𝜃


+ + + 𝜌𝑓𝑟 = 𝜌𝑎𝑟
𝜕𝑟 𝑟 𝜕𝜃 𝑟

𝜕𝜎𝜃𝑟 1 𝜕𝜎𝜃𝜃 2𝜎𝑟𝜃


+ + + 𝜌𝑓𝜃 = 𝜌𝑎𝜃
𝜕𝑟 𝑟 𝜕𝜃 𝑟

Strain-displacement relationships:

𝜕𝑢 1 𝜕𝑣 1 𝜕𝑢 𝑣
𝜀𝑟𝑟 = 𝜀𝑟𝜃 = ( + − )
𝜕𝑟 2 𝜕𝑟 𝑟 𝜕𝜃 𝑟
1 𝜕𝑣 𝑢 1 1 𝜕𝑤 𝜕𝑣
Normal: 𝜀𝜃𝜃 = 𝑟 𝜕𝜃 + 𝑟 Shear: 𝜀𝜃𝑧 = 2 (𝑟 𝜕𝜃 + 𝜕𝑧 )
𝜕𝑤 1 𝜕𝑢 𝜕𝑤
{ 𝜀𝑧𝑧 = { 𝜀𝑧𝑟 = 2 ( 𝜕𝑧 + )
𝜕𝑧 𝜕𝑥

Hooke’s law:

𝐸
𝜎𝑟𝑟 = [(1 − 𝜈)𝜀𝑟𝑟 + 𝜈(𝜀𝜃𝜃 + 𝜀𝑧𝑧 )]
(1 + 𝜈)(1 − 2𝜈)

15
MECH3361/9361 Semester 2, 2016

𝐸
𝜎𝜃𝜃 = [(1 − 𝜈)𝜀𝜃𝜃 + 𝜈(𝜀𝑧𝑧 + 𝜀𝑟𝑟 )]
(1 + 𝜈)(1 − 2𝜈)

𝐸
𝜎𝑧𝑧 = [(1 − 𝜈)𝜀𝑧𝑧 + 𝜈(𝜀𝑟𝑟 + 𝜀𝜃𝜃 )]
(1 + 𝜈)(1 − 2𝜈)

𝐸
𝜎𝑧𝑟 = 𝜀
1 + 𝜈 𝑧𝑟

4.8.2. Solution using the displacement method


Step 1: Check the boundary conditions

At 𝑟 = 𝑅𝑖 ∶ 𝜎𝑟𝑟 = −𝑝𝑖 , 𝜎𝑟𝜃 = 0, 𝜎𝑟𝑧 = 0

At 𝑟 = 𝑅𝑜 ∶ 𝜎𝑟𝑟 = −𝑝𝑜 , 𝜎𝑟𝜃 = 0, 𝜎𝑟𝑧 = 0

Step 2: Analysis

The deformation is axisymmetric and under plane strain, so the deformation is independent
of coordinates 𝜃 and 𝑧. Thus, the circumferential and axial displacements, 𝑣 and 𝑤, will
vanish, and the displacements can be expressed as:

𝑢 = 𝑢(𝑟)
𝑣=0
𝑤=0

Step 3: Strain-displacement relationships

𝜕𝑢
𝜀𝑟𝑟 =
𝜕𝑟
1 𝜕𝑣 𝑢 1 𝑢 𝑢
𝜀𝜃𝜃 = + = 0+ =
𝑟 𝜕𝜃 𝑟 𝑟 𝑟 𝑟
𝜕𝑤
𝜀𝑧𝑧 = =0
𝜕𝑧

1 𝜕𝑣 1 𝜕𝑢 𝑣 1 𝜕𝑢(𝑟) 1
𝜀𝑟𝜃 = ( + − )= ( )= 0=0
2 𝜕𝑟 𝑟 𝜕𝜃 𝑟 2𝑟 𝜕𝜃 2𝑟
1 1 𝜕𝑤 𝜕𝑣 1 1
𝜀𝜃𝑧 = ( + ) = ( 0 + 0) = 0
2 𝑟 𝜕𝜃 𝜕𝑧 2 𝑟
1 𝜕𝑢 𝜕𝑤 1 𝜕𝑢(𝑟) 1
𝜀𝑧𝑟 = ( + )= ( + 0) = (0 + 0) = 0
2 𝜕𝑧 𝜕𝑥 2 𝜕𝑧 2

Step 4: Apply Hooke’s law

𝐸 𝜕𝑢 𝑢
𝜎𝑟𝑟 = [(1 − 𝜈) + 𝜈 ( + 0)]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟

16
MECH3361/9361 Semester 2, 2016

𝐸 𝜕𝑢 𝑢
= [(1 − 𝜈) +𝜈 ]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟

𝐸 𝑢 𝜕𝑢
𝜎𝜃𝜃 = [(1 − 𝜈) + 𝜈 (0 + )]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝜕𝑟
𝐸 𝑢 𝜕𝑢
= [(1 − 𝜈) + 𝜈 ]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝜕𝑟

𝐸 𝜕𝑢 𝑢
𝜎𝑧𝑧 = [(1 − 𝜈)0 + 𝜈 ( + )]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
𝐸𝜈 𝜕𝑢 𝑢
= [ + ]
(1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟

𝐸 𝐸
𝜎𝑧𝑟 = 𝜀𝑧𝑟 = 0=0
1+𝜈 1+𝜈

Step 5: Equilibrium equations

For this question, we will assume that body forces are negligible and there is no acceleration.
Therefore, the equations are:

𝜕𝜎𝑟𝑟 1 𝜕𝜎𝑟𝜃 ∂𝜎rz 𝜎𝑟𝑟 − 𝜎𝜃𝜃


+ + + +0 = 0
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

𝜕𝜎𝜃𝑟 1 𝜕𝜎𝜃𝜃 ∂𝜎𝜃z 2𝜎𝑟𝜃


+ + + +0 =0
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

𝜕𝜎𝑧𝑟 1 𝜕𝜎𝑧𝜃 ∂𝜎zz 𝜎𝑟𝑧


+ + + +0=0
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

The second and third equations are satisfied automatically:

𝜕𝜎𝜃𝑟 1 𝜕𝜎𝜃𝜃 ∂𝜎𝜃z 2𝜎𝑟𝜃 𝜕0 1 𝜕𝜎𝜃𝜃 (𝑟) ∂0 2(0)


+ + + = + + + =0
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟 𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

𝜕𝜎𝑧𝑟 1 𝜕𝜎𝑧𝜃 ∂𝜎zz 𝜎𝑟𝑧 𝜕0 1 𝜕0 ∂𝜎zz (𝑟) 0


+ + + = + + + =0
𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟 𝜕𝑟 𝑟 𝜕𝜃 ∂z 𝑟

For the first equation, since there are no shear stresses:

𝜕𝜎𝑟𝑟 𝜎𝑟𝑟 − 𝜎𝜃𝜃


+ =0
𝜕𝑟 𝑟

Now we can substitute Hooke’s law:

17
MECH3361/9361 Semester 2, 2016

𝜕 𝐸 𝜕𝑢 𝑢
𝐿𝐻𝑆 = { [(1 − 𝜈) + 𝜈 ]}
𝜕𝑟 (1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
1 𝐸 𝜕𝑢 𝑢
+ { [(1 − 𝜈) +𝜈 ]
𝑟 (1 + 𝜈)(1 − 2𝜈) 𝜕𝑟 𝑟
𝐸 𝑢 𝜕𝑢
− [(1 − 𝜈) + 𝜈 ]}
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝜕𝑟
𝜕 𝜕𝑢 𝑢 1 𝜕𝑢 𝑢 𝑢 𝜕𝑢
= {(1 − 𝜈) + 𝜈 } + {[(1 − 𝜈) + 𝜈 ] − [(1 − 𝜈) + 𝜈 ]}
𝜕𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝑟
𝜕 𝜕𝑢 1 1 𝜕𝑢 𝜕𝑢 𝑢 𝑢
= {(1 − 𝜈) + 𝜈 ( ) 𝑢} + { − 2𝜈 − + 2𝜈 }
𝜕𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝑟
2
𝜕 𝑢 𝑢 𝜈 𝜕𝑢 1 𝜕𝑢 2𝜈 𝜕𝑢 𝑢 𝑢
= {(1 − 𝜈) 2 + (−𝜈 2 + )} + { − − 2 + 2𝜈 2 }
𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝑟
𝜕2𝑢 𝑢 𝜈 𝜕𝑢 1 𝜕𝑢 2𝜈 𝜕𝑢 𝑢 𝑢
= (1 − 𝜈) 2 − 𝜈 2 + + − − 2 + 2𝜈 2
𝜕𝑟 𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝑟
𝜕2𝑢 1 𝜕𝑢 𝑢
= (1 − 𝜈) 2 + (1 − 𝜈) − (1 − 𝜈) 2
𝜕𝑟 𝑟 𝜕𝑟 𝑟
𝜕 2 𝑢 1 𝜕𝑢 𝑢
= 2+ −
𝜕𝑟 𝑟 𝜕𝑟 𝑟 2
=0

Step 6: Solve the above linear and static ordinary differential equation

1
We assume a trial solution 𝑢 = 𝐶1 𝑟 + 𝐶2 𝑟, in which 𝐶1 and 𝐶2 are constants to be determined
by using the BCs.

Step 7: Plug this trial function into the strain-displacement relations

𝜕𝑢 𝜕 1 1
𝜀𝑟𝑟 = = (𝐶1 𝑟 + 𝐶2 ) = 𝐶1 − 𝐶2 2
𝜕𝑟 𝜕𝑟 𝑟 𝑟

1 𝜕𝑣 𝑢 𝑢 1 1 1
𝜀𝜃𝜃 = + = = (𝐶1 𝑟 + 𝐶2 ) = 𝐶1 + 𝐶2 2
𝑟 𝜕𝜃 𝑟 𝑟 𝑟 𝑟 𝑟

𝜀𝑧𝑟 = 𝜀𝑧𝜃 = 𝜀𝑧𝑧 = 0

Substituting this into Hooke’s law gives:

𝐸
𝜎𝑟𝑟 = [(1 − 𝜈)𝜀𝑟𝑟 + 𝜈(𝜀𝜃𝜃 + 𝜀𝑧𝑧 )]
(1 + 𝜈)(1 − 2𝜈)
𝐸 1 1
= [(1 − 𝜈) (𝐶1 − 𝐶2 2 ) + 𝜈 (𝐶1 + 𝐶2 2 )]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝑟
𝐸 1 𝜈 𝜈
= [𝐶1 − 𝐶2 2 − 𝐶1 𝜈 + 𝐶2 2 + 𝐶1 𝜈 + 𝐶2 2 ]
(1 + 𝜈)(1 − 2𝜈) 𝑟 𝑟 𝑟
𝐸 𝐶2
= [𝐶 − (1 − 2𝜈)]
(1 + 𝜈)(1 − 2𝜈) 1 𝑟 2

18
MECH3361/9361 Semester 2, 2016

𝐸𝐶1 1 𝐸𝐶2
=[ ]− 2[ ]
(1 + 𝜈)(1 − 2𝜈) 𝑟 (1 + 𝜈)
𝐵
=𝐴− 2
𝑟
𝐸𝐶
1 2𝐸𝐶
where 𝐴 = [(1+𝜈)(1−2𝜈)] and 𝐵 = [(1+𝜈)].

Similarly,

𝐵
𝜎𝜃𝜃 = 𝐴 +
𝑟2

Now we need to determine 𝐴 and 𝐵.

(Note that since we incorporated all of the equations in our solution, it should satisfy all of
them.)

Step 8: Apply BCs to determine the constants

𝐵
𝜎𝑟𝑟 (𝑟 = 𝑅𝑖 ) = 𝐴 − = −𝑝𝑖
𝑅𝑖 2

𝐵
𝜎𝑟𝑟 (𝑟 = 𝑅𝑜 ) = 𝐴 − = −𝑝𝑜
𝑅𝑜 2

Solving these two equations simultaneously leads to:

𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝐴=
𝑅𝑜2 − 𝑅𝑖2

(𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2


𝐵=
𝑅𝑜2 − 𝑅𝑖2

Therefore,

(1 + 𝜈)(1 − 2𝜈) (1 + 𝜈)(1 − 2𝜈) 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2


𝐶1 = 𝐴= [ ]
𝐸 𝐸 𝑅𝑜2 − 𝑅𝑖2

1+𝜈 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2


𝐶2 = 𝐵= [ ]
𝐸 𝐸 𝑅𝑜2 − 𝑅𝑖2

Step 9: Calculate all the functions (for the 15 unknowns)

19
MECH3361/9361 Semester 2, 2016

Displacements:

1 (1 + 𝜈)(1 − 2𝜈) 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝑢 = 𝐶1 𝑟 + 𝐶2 = [ ]𝑟 + [ ]
𝑟 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝑣=0

𝑤=0

Strains:

1 (1 + 𝜈)(1 − 2𝜈) 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜀𝑟𝑟 = 𝐶1 − 𝐶2 2 = [ ]− [ ] 2
𝑟 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝑟

1 (1 + 𝜈)(1 − 2𝜈) 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜀𝜃𝜃 = 𝐶1 + 𝐶2 = [ ] + [ ] 2
𝑟2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝜀𝑧𝑧 = 0

𝜀𝑟𝜃 = 𝜀𝜃𝑧 = 𝜀𝑧𝑟 = 0

Stresses:

𝐵 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜎𝑟𝑟 (𝑟) = 𝐴 − = − ( 2)
𝑟2 𝑅𝑜2 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝐵 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜎𝜃𝜃 (𝑟) = 𝐴 + 2 = + ( 2)
𝑟 𝑅𝑜2 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝜎𝑧𝑧 (𝑟) = 2𝜈 ( )
𝑅𝑜2 − 𝑅𝑖2

𝜎𝑟𝜃 = 𝜎𝜃𝑧 = 𝜎𝑧𝑟 = 0

4.8.3. Plane stress formulation

Plane strain Plane stress

Fig. 4.10 Pressurised cylinder with plane strain and plane stress

𝐸(1+2𝜈) 𝜈
Replacing 𝐸 and 𝜈 by 𝐸 ∗ = (1+𝜈)2
and 𝜈 ∗ = respectively (recall Section 4.5.4), we can
1+𝜈
further obtain the solution to the corresponding plane stress problem (illustrated above).

20
MECH3361/9361 Semester 2, 2016

(1 + 𝜈 ∗ )(1 − 2𝜈 ∗ ) 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 ∗ (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝑢𝑝−𝑠𝑡𝑟𝑎𝑖𝑛 = [ ]𝑟 + [ ]
𝐸∗ 𝑅𝑜2 − 𝑅𝑖2 𝐸∗ 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝜈 𝜈 𝜈
[1 + ( )] [1 − 2 ( )] 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1+( ) (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1
𝑢𝑝−𝑠𝑡𝑟𝑒𝑠𝑠 = 1 + 𝜈 1 + 𝜈 [ ]𝑟 + 1 + 𝜈 [ ]
[
𝐸(1 + 2𝜈)
] 𝑅𝑜2 − 𝑅𝑖2 [
𝐸 (1 + 2𝜈)
] 𝑅𝑜2 − 𝑅𝑖2 𝑟
(1 + 𝜈)2 (1 + 𝜈)2
1 + 2𝜈 1 − 𝜈 1 + 2𝜈
[ ][ ] 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 [ ] (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1
= 1 + 𝜈 1 + 𝜈 [ ]𝑟 + 1 + 𝜈 [ ]
[
𝐸 (1 + 2𝜈 )
] 𝑅𝑜2 − 𝑅𝑖2 [
𝐸(1 + 2𝜈)
] 𝑅𝑜2 − 𝑅𝑖2 𝑟
(1 + 𝜈)2 (1 + 𝜈)2

Displacements:

1 − 𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝑢 =( )[ ]𝑟 + ( )[ ]
𝐸 𝑅𝑜2 − 𝑅𝑖2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝑣=0

2𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝑤=− [ ]𝑧 + 𝐶
𝐸 𝑅𝑜2 − 𝑅𝑖2

Strains:

1 − 𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜀𝑟𝑟 = ( )[ ] − ( )[ ] 2
𝐸 𝑅𝑜2 − 𝑅𝑖2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝑟

1 − 𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 1 + 𝜈 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜀𝜃𝜃 =( )[ ]+ [ ] 2
𝐸 𝑅𝑜2 − 𝑅𝑖2 𝐸 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝜈 𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2
𝜀𝑧𝑧 =− [ ]
𝐸 𝑅𝑜2 − 𝑅𝑖2

𝜀𝑟𝜃 = 𝜀𝜃𝑧 = 𝜀𝑧𝑟 = 0

Stresses:

𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜎𝑟𝑟 (𝑟) = − ( 2)
𝑅𝑜2 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝑝𝑖 𝑅𝑖2 − 𝑝𝑜 𝑅𝑜2 (𝑝𝑖 − 𝑝𝑜 )𝑅𝑖2 𝑅𝑜2 1


𝜎𝜃𝜃 (𝑟) = + ( 2)
𝑅𝑜2 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟

𝜎𝑧𝑧 (𝑟) = 0

𝜎𝑟𝜃 = 𝜎𝜃𝑧 = 𝜎𝑧𝑟 = 0

21
MECH3361/9361 Semester 2, 2016

4.8.4. Remarks
𝜎𝑟𝑟 and 𝜎𝜃𝜃 are independent of material properties. Thus, any cylinder of the same
dimensions will experience the same stress values. If strength is a major concern, one
should simply select the material with the highest strength.
However, the displacements and strains are dependent on material properties. If
stiffness is a major concern, a higher Young’s modulus material should be chosen.
When 𝑝𝑜 = 0, we have:
𝑝𝑖 𝑅𝑖2 𝑝𝑖 𝑅𝑖2 𝑅𝑜2 1
𝜎𝑟𝑟 (𝑟) = 2 − ( )
𝑅𝑜 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟 2
𝑝𝑖 𝑅𝑖2 𝑝𝑖 𝑅𝑖2 𝑅𝑜2 1
𝜎𝜃𝜃 (𝑟) = 2 + ( )
{ 𝑅𝑜 − 𝑅𝑖2 𝑅𝑜2 − 𝑅𝑖2 𝑟 2
Since 𝑟 ≤ 𝑅𝑜 , the radial stress 𝜎𝑟𝑟 ≤ 0 (always negative) and 𝜎𝜃𝜃 ≥ 0 (always
positive). Thus, 𝜎𝑟𝑟 ≤ 𝜎𝑧𝑧 (= 0) ≤ 𝜎𝜃𝜃 . Since all shear stresses are zero, the principle
stresses are:
𝜎1 = 𝜎𝜃𝜃 , 𝜎2 = 𝜎𝑧𝑧 = 0, 𝜎3 = 𝜎𝑟𝑟

4.9. Saint-Venant’s principle


4.9.1. Definition
In the cantilever beam problem, there are differences in stress contours between the analytical
and FEA solution methods, as shown in Fig. 4.11.

y P

0 x

FEA

Theoretical solution
Fig. 4.11 Saint Venant Principles

Saint Venant observed that under pure bending, a beam conforms to a rigorous solution only
when the external forces applied at the ends of beams are distributed over the end is the same
as internal stress distribution, i.e. linear distribution.

22
MECH3361/9361 Semester 2, 2016

Saint Venant’s Principle:


If the forces acting on a small portion of the surface of an elastic
body are replaced by another statically equivalent system of forces
acting on the same portion of that surface, such redistribution of
loading produces substantial change in stress locally but has
negligible effect on the stresses at distances that are far in
comparison with the dimensions of the loaded surface.

There are two key assumptions here:

1. The loading area is very small relative to the overall dimensions. This means the
affected area will be much smaller than the unaffected area. For instance, in the
tensile bar in Fig 4.12, 𝐿 ≫ 𝑎, so the affected zone will be roughly |𝒛| ≈ 𝒂.
Affected Affected
unaffected zone
zone zone

a
z z

Fig. 4.12 Affected and unaffected areas


2. Force replacement must be statically equivalent, i.e. the replacement must not
change the resultant force nor the resultant couple. For the bar above, loading may
occur on both ends, be distributed over multiple points, or perhaps one end may be
fixed, but so long as the resultants are identical, the approximate affected and
unaffected zones are still valid as marked.

4.9.2. Examples

Tensile testing

F Elastic Yielding Hardening Necking

Ultimate
stress u
Standard
Specimen Yield stress Y Fracture
Proportional pl stress f
limit

Plastic behavior
F
Elastic
behavior

When performing a tensile test of a standard material specimen, the way the specimen is
clamped should have no effect on the stress and deformation in the centre regions. Testing

23
MECH3361/9361 Semester 2, 2016

procedures require a minimum specimen length in order to avoid the end effect on the testing
result. This is an application of Saint-Venant’s principle.

Four-point bending

M(x)

Affect zone
Strain gauge testing

worse
position Better
position

For more stable and reliable testing results, strain gauges should be placed in the far field as
shown above. Strain gauges that are too close to a loading point may yield measurements that
affected by the local stress distribution.

Cantilever beams

-1000
-2000N/m
-1000 -2000

The FEA result above shows the stress distributions when end forces are applied in different
ways. From the contours, it is clear that only a small region of the beam nearest to the loading
surface is affected. Stresses in the far field are comparable in all three cases illustrated.

Example 4.4
A solid circular shaft in a machine, as shown in the figure below, is subjected to tension by a
concentrated force P, torsion by torque Q, and bending by bending moment M in the xz-plane
(i.e. 𝑀 = 𝑀𝑦𝑦 ). The radius of the shaft is R and the length is L (where 𝐿 ≫ 𝑅). Describe the
boundary conditions using Saint Venant’s principle.

x x
Q
M M
dr
d rd y
P z
P
Q
y

24
MECH3361/9361 Semester 2, 2016

Solution

Although we know the resultant force P, M and T on the ends of shaft, we do not know the
exact stress distributions that make up these resultant forces. Since 𝐿 ≫ 𝑅, Saint-Venant’s
principle means we can express these stress BCs using statically equivalent conditions (i.e.
independent of the unknown detailed stress distribution).

The BCs can be expressed as follows:

At 𝑧 = 0 and 𝑧 = −𝐿, for 0 ≤ 𝑟 ≤ 𝑅 and 0 ≤ 𝜃 ≤ 2𝜋 (the flat left and right end faces):

𝑑𝐴 = (𝑟 𝑑𝜃)(𝑑𝑟)
2𝜋 𝑅 2𝜋 𝑅

𝑃 = ∫ ∫ 𝜎𝑧𝑧 𝑑𝐴 = ∫ ∫ 𝜎𝑧𝑧 𝑟 𝑑𝑟 𝑑𝜃
0 0 0 0

2𝜋 𝑅 2𝜋 𝑅

𝑀𝑥𝑥 = ∫ ∫[𝜎𝑧𝑧 𝑑𝐴][𝑟𝑠𝑖𝑛𝜃] = ∫ ∫ 𝑟 2 𝜎𝑧𝑧 𝑠𝑖𝑛𝜃 𝑑𝑟 𝑑𝜃 = 0


0 0 0 0

2𝜋 𝑅 2𝜋 𝑅

𝑀𝑦𝑦 = ∫ ∫[𝜎𝑧𝑧 𝑑𝐴][𝑟𝑐𝑜𝑠𝜃] = ∫ ∫ 𝑟 2 𝜎𝑧𝑧 𝑐𝑜𝑠𝜃 𝑑𝑟 𝑑𝜃 = 𝑀


0 0 0 0

2𝜋 𝑅 2𝜋 𝑅

𝑀𝑧𝑧 = ∫ ∫[𝜎𝑧𝜃 𝑑𝐴][𝑟] = ∫ ∫ 𝑟 2 𝜎𝑧𝜃 𝑑𝑟 𝑑𝜃 = −𝑄


0 0 0 0

𝜎𝑧𝑟 = 0

At 𝑟 = 𝑅, 0 ≤ 𝜃 ≤ 2𝜋, −𝐿 ≤ 𝑧 ≤ 0 (cylindrical surface):

𝜎𝑟𝑟 = 𝜎𝑟𝜃 = 𝜎𝑟𝑧 = 0

25

You might also like